SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2506.02493 · CVPR · 2025

Towards In-the-wild 3D Plane Reconstruction from a Single Image

Jiachen Liu, Rui Yu, Sharon Huang, Hengkai Guo, Sili Chen

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 9 functions out of this paper's own repositories and ran 4 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

RepositoryRoleRan
jcliu0428/ZeroPlane canonical 4 of 9
FunctionStatusWhere it lives
conv3x3 Ran jcliu0428/ZeroPlane/ZeroPlane/modeling/backbone/hrnet.py
code served (permissive licence) · get_code("fac5364e2f53c6db")
l1_loss Ran jcliu0428/ZeroPlane/ZeroPlane/modeling/criterion.py
code served (permissive licence) · get_code("2cde99938ddc050b")
window_partition Ran jcliu0428/ZeroPlane/ZeroPlane/modeling/backbone/swin.py
code served (permissive licence) · get_code("f9fd6241d935f07b")
window_reverse Ran jcliu0428/ZeroPlane/ZeroPlane/modeling/backbone/swin.py
code served (permissive licence) · get_code("fb32094c6dbece71")
batch_dice_loss Not yet run jcliu0428/ZeroPlane/ZeroPlane/modeling/matcher.py
code served (permissive licence) · get_code("bc2cb481a75c370d")
batch_sigmoid_ce_loss Not yet run jcliu0428/ZeroPlane/ZeroPlane/modeling/matcher.py
code served (permissive licence) · get_code("1edd24985036b0bf")
build_hrnet Not yet run jcliu0428/ZeroPlane/ZeroPlane/modeling/backbone/hrnet.py
code served (permissive licence) · get_code("0b5d28998e36d4d7")
dice_loss Not yet run jcliu0428/ZeroPlane/ZeroPlane/modeling/criterion.py
code served (permissive licence) · get_code("89f75e54ff128be0")
sigmoid_ce_loss Not yet run jcliu0428/ZeroPlane/ZeroPlane/modeling/criterion.py
code served (permissive licence) · get_code("d0c61e8dba511aa3")

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

3D plane reconstruction from a single image is a crucial yet challenging topic in 3D computer vision. Previous stateof-the-art (SOTA) methods have focused on training their system on a single dataset from either indoor or outdoor domain, limiting their generalizability across diverse testing data. In this work, we introduce a novel framework dubbed ZeroPlane, a Transformer-based model targeting zero-shot 3D plane detection and reconstruction from a single image, over diverse domains and environments. To enable datadriven models across multiple domains, we have curated a large-scale planar benchmark, comprising over 14 datasets and 560,000 high-resolution, dense planar annotations for diverse indoor and outdoor scenes. To address the challenge of achieving desirable planar geometry on multi-dataset training, we propose to disentangle the representation of plane normal and offset, and employ an exemplar-guided, classification-then-regression paradigm to learn plane and offset respectively. Additionally, we employ advanced backbones as image encoder, and present an effective pixelgeometry-enhanced plane embedding module to further facilitate planar reconstruction. Extensive experiments across multiple zero-shot evaluation datasets have demonstrated that our approach significantly outperforms previous methods on both reconstruction accuracy and generalizability, especially over in-the-wild data. Our code and data are available at: https://github.com/jcliu0428/ZeroPlane.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2506.02493")
get_code_for_paper("2506.02493")
have("2506.02493")

Connect an agent — have() is free.